Domain-Agnostic Structured Search Query Exploration

1. A method for a search engine to provide domain-agnostic options for structured exploration based on a query, comprising: clustering a corpus of historical queries into a plurality of query clusters, each query cluster of the plurality being clustered to include historical queries demonstrating a common query structure for that query cluster, and each query structure including a variable query term; for each query cluster of the plurality, identifying a set of variable query terms in the historical queries; receiving a new user query; identifying a query structure of the new user query; selecting a specific query cluster matching the query structure of the new user query; identifying the new variable query term in the query structure of the new user query; and providing a query exploration interface offering a plurality of different user-selectable options for structured exploration based on the new user query, each of the plurality of different user-selectable options corresponding to a particular variable query term in the identified set of variable query terms for the specific query cluster identified as matching the query structure of the new user query.

2. The method of claim 1 , wherein the common query structure for a query cluster corresponds to a query topic common to historical queries in the query cluster, wherein each historical query in the query cluster corresponds to a combination of the query topic and a corresponding specific query term in place of the new variable query term.

3. The method of claim 1 , wherein the specific query cluster is one of a plurality of candidate query clusters, each candidate query cluster demonstrating a candidate query structure matching the query structure of the new user query, the method further comprising assessing a correspondence between the new user query and historical queries of each candidate query cluster, and the specific query cluster having a highest correspondence.

4. The method of claim 1 , further comprising selecting a second specific query cluster, having a second common query structure matching a second query structure of the new user query, and providing a second query exploration interface offering a second plurality of different user-selectable options for structured exploration based on the new user query based on the second specific query cluster.

5. The method of claim 4 , further comprising selecting a third specific query cluster, having a third common query structure matching a third query structure of the new user query, and providing a third query exploration interface offering a third plurality of different user-selectable options for structured exploration based on the new user query based on the third specific query cluster, wherein the common query structure, the second common query structure, and the third common query structure each correspond to a topic associated with the new user query, and different user-selectable options of each exploration interface are associated with different queries related to a corresponding topic.

6. The method of claim 1 , wherein the new user query is a natural language query including an ordered plurality of words, and the query structure includes a subsequence of the ordered plurality of words.

7. The method of claim 1 , wherein the new user query is a computer-readable query command and the query structure includes a syntactic feature of the computer-readable query command.

8. The method of claim 1 , wherein the new user query matches the query structure while leaving the new variable query term unspecified, and wherein the plurality of different user-selectable options in the query exploration interface are associated with refined queries, each refined query based on specifying a different suggested query term for the new variable query term.

9. The method of claim 1 , wherein the new user query includes the query structure and specifies a specific query term for the new variable query term, and wherein the plurality of different user-selectable options in the query exploration interface are associated with alternative queries, each alternative query based on replacing the specific query term with a different alternative query term.

10. The method of claim 1 , further comprising determining the common query structure for a query cluster based on identifying a most popular phrase occurring in a plurality of queries in the query cluster.

11. The method of claim 1 , further comprising determining the common query structure for a query cluster based on identifying a plurality of contextually-related entities mentioned in historical queries of the query cluster.

12. The method of claim 1 , wherein clustering the corpus of historical queries into a plurality of query clusters includes, responsive to an initial query cluster having more than a threshold number of queries, multi-level clustering the initial query cluster into a plurality of sub-clusters.

13. The method of claim 1 , further comprising determining the common query structure for a query cluster based on multi-level clustering a plurality of historical queries of the query cluster into sub-clusters based on a contextual relationship between entities mentioned in the plurality of historical queries of the query cluster.

14. The method of claim 13 , wherein the contextual relationship between entities is determined based on comparing semantic vector encodings for entities mentioned in the plurality of historical queries of the query cluster.

15. The method of claim 13 , wherein the contextual relationship between entities is determined based on a comparison among search context vectors derived from web search result listings associated with historical queries of the query cluster.

16. The method of claim 13 , further comprising assessing entity co-occurrence vector data indicating whether each entity is mentioned by each document of a corpus of documents, wherein the contextual relationship between entities is determined based on comparing the entity co-occurrence vector data for a plurality of entities mentioned in historical queries in the query cluster.

17. The method of claim 16 , wherein the corpus of documents includes a plurality of web tables, each web table associated with an index of entities, and the entity co-occurrence vector data is based on an inverted index indicating, for each entity, a listing of each web table with an associated index including the entity.

18. The method of claim 16 , wherein the entity co-occurrence vector data is assessed selectively for different pairs of entities, including selectively assessing the entity co-occurrence vector data solely with regard to pairs of entities satisfying a hash collision property.

19. A computing system, comprising: a logic subsystem; a communication subsystem; and a storage device holding instructions executable by the logic device to: cluster a corpus of historical queries into a plurality of query clusters, each query cluster of the plurality being clustered to include historical queries demonstrating a common query structure for that query cluster, and each query structure including a variable query term; for each query cluster of the plurality, identify a set of variable query terms in the historical queries; receive a new user query, from a user device via the communication subsystem; identify a query structure of the new user query; select a specific query cluster matching the query structure of the new user query; identify a new variable query term in the query structure of the new user query; and provide, to the user device via the communication subsystem, a query exploration interface offering a plurality of different user-selectable options for structured exploration based on the new user query, each of the plurality of different user-selectable options corresponding to a particular variable query term in the identified set of variable query terms for the specific query cluster identified as matching the query structure of the new user query.

20. A method for a search engine to provide domain-agnostic options for structured exploration based on a query, comprising: clustering a corpus of historical queries into a plurality of query clusters, each query cluster of the plurality being clustered to include historical queries demonstrating a common query structure for that query cluster, and each query structure including a variable query term; for each query cluster of the plurality, identifying a set of variable query terms in the historical queries; receiving a new user query; identifying a first query structure of the new user query and a second, different query structure of the new user query; selecting a first query cluster matching the first query structure of the new user query, and selecting a second query cluster matching the second query structure of the new user query; identifying a first variable query term corresponding to the first query structure, and identifying a second variable query term corresponding to the second query structure; and providing a search interface including a first plurality of different user-selectable options for replacing the first variable query term, and a second plurality of different user-selectable options for replacing the second variable query term, wherein: a first user-selectable option of the search interface is selectable to cause the search interface to provide results for a first exploratory query based on the new user query with the first variable query term replaced based on the first user-selectable option; and a second user-selectable option of the search interface is further selectable to cause the search interface to provide results for a second, further exploratory query based on the new user query with the second variable query term replaced based on the second user-selectable option.